Study Of Analysis And Data Mining On Chinese Medicine Patent

Patent refers to the invention, which is protected by the law, including the invention, the utility model, the designs. If applicants submit applications to the patent administrative departments, they can enjoy the ownership and ascendancy of the invention within specified region and period after disclosure and approval. The information within the public document of application process is called patent information, which is a complex strategic resource with legal, technical, economic information. Regularly arrangement and analysis on drug patent make researchers are able to accurately grasp the market and technology situation, and carry out researches and development purposefully.In this study, with the help of computer aided, Chinese medicine patent information in specific field is analysed from multi angle, including history, development, trend and pattern and compatibility of medicines. And a series of processes and methods to analyse patent information is summarized. In order to provide some references about the way to process and analyse Chinese medicine patent. Main technical key points and steps are:the choice of the source database, retrieval strategy adjustment, and supplement and clean of patent information, demand analysis, patent data transform and statistics, using of text mining algorithm.In this paper, there are two patent information sources:World Traditional Medicine Patent Database developed by Beijing Eastlinden Technology Co., Ltd. And Espacenet developed by European Patent Office. The former is a professional Chinese medicine patent database, and the later is a worldwide patent database that free access to the database of over 90 million patents. With the help of these two databases we can ensure the reliability of retrieved data.Retrieved results is processed using Python, R, SQLite3 and their third party extensions, such as SciPy, NumPy, MatPlotLib, Gensim, Pheatmap and so on. In addition to the common methods of statistical analysis, the text mining algorithms are practice on Chinese medicine patent data, including keyword extraction, latent semantic analysis and others.Through the analysis of patent about 13 kinds of herbs (come from 11 kinds of medicinal plants),5 functions and 1 enterprise, a set of conclusions are summarized. Among the Chinese herbal plants researched by this paper, the five most potential medicinal plants for development are yam, medlar, asarum, peony, and radix codonopsis. Chinese mainland’s patent family accounted for an absolute advantage, but the output is relatively low. The region which is most attractive to applicants is Chinese mainland. There are 6 kinds of patent applications in the field of traditional Chinese medicine: pharmaceutical companies, cosmetic companies, food companies, biotechnology companies, research institutes and universities, intellectual property services enterprises. Of which, pharmaceutical companies own the most patent families, and the continuity of scientific research’s patent family is better. Patent information, technical services are rising in the Chinese medicine industry. In addition, the law of compatibility of different medicinal herbs are summarized.